Promo code optimization model

The company i work for as data analyst relies heavily on Promo codes (magazines, catalogs & online) to make sales.

Currently, discount rates are chosen randomly and intuitively. Any recommendations on which direction to look in terms of models and processes to able to optimize the coupon rates ?

Huge amount of historical data is available(coupon rates, campaign ROI & the customer segment that best responded to a specific coupon.)

analyst code data data analyst datascience historical data look magazines optimization processes recommendations sales terms work